Rapidly scaling online storage to serve over 1 billion ChatGPT users (Part II)
Learn how OpenAI scaled Habitat with reliable multi-tenancy, global caching, and Azure Cosmos DB to deliver 12× faster reads for ChatGPT.
Search headlines, summaries and articles to connect the dots.
Learn how OpenAI scaled Habitat with reliable multi-tenancy, global caching, and Azure Cosmos DB to deliver 12× faster reads for ChatGPT.
OpenAI explains how we strengthened containment for frontier model training and evaluation, including offline environments and high-compute red-teaming.
As AI transitions from the experimental stage to large-scale production, corporate competitiveness is shifting from model selection to the serving harness—a suite of engineering systems that encompasses architecture design, memory management, scheduling strategies, and inference optimization. In an article published in MIT Technology Review, EXL points out that the key determinant of AI’s real-world impact is no longer model parameters or benchmark scores, but rather the ability to translate model capabilities into business value with controllable costs, reliable latency, and sustainable scalability.
Anthropic has officially launched Claude Duet—an AI agent application for professionals and small businesses, built on the Claude Opus 5.5 model. It supports over 6,000 connectors, can automatically generate daily briefs, build dashboards, suggest actions, and perform background tasks; currently, it is available only to Pro and Max subscribers on a waitlist for trial use.
Celonis Chief Scientist Devavrat Shah argues that enterprise AI must be embedded in real-world business contexts—such as workflows, role-based permissions, and process constraints—in order to achieve “decision intelligence”; AI divorced from context is prone to errors, resource waste, or ineffective outputs. His technical approach relies on a “context layer” and a digital twin architecture, emphasizing incremental validation, role-based governance, and workforce empowerment. However, all these claims are subjective assertions and have not been independently verified.
At the GTC Berlin conference, European financial institutions including Revolut, Starling Bank, Lloyds, and Santander announced that they are building production‑grade AI capabilities based on NVIDIA’s accelerated computing platform and open‑source weight models—such as PRAGMA, the Nemotron series, and the Kumo family—covering core use cases like trading analytics, speech recognition, fraud detection, and data pipeline acceleration, while emphasizing data sovereignty, cost control, and compliant deployment.
OpenReplay replaced Postgres with ClickHouse to ingest billions of events and up to 6 TB a day on a single, vertically scaled instance, with sub-second queries and retention handled by a single TTL setting.
Connect Cursor to Langfuse to trace every agent turn, pinpoint expensive model and tool calls, and use evaluations to confirm a fix actually helps, with ClickHouse powering the analytics underneath.
AWS Solutions Architecture Team has launched Deep Insight, an open-source multi-agent system designed for automated CSV data analysis and natural-language report generation. It systematically addresses five common challenges in production environments—agent orchestration, role-based model selection, stable operation, secure code execution, and observability—and has been validated in real-world scenarios such as LG Electronics and the re:Invent 2025 TechRecon.
Claude Mods is Anthropic’s programmable plugin system for Claude Code, representing an early form of the “Attention Interface”—it shifts the traditional machine-oriented agent harness toward a customizable interaction layer centered on human attention. Users can write Mods in TypeScript/JS or have Claude generate them automatically using natural language, thereby determining when to be interrupted, what content to see, and what decisions to make.
Cloudflare uses OpenAI Daybreak to investigate vulnerabilities and propose fixes for review. Its developers use Codex to build faster and improve code quality.
While foundation models and vision transformers (ViTs) have redefined visual representations, current active learning (AL) strategies for these backbones largely operate at the patch or region level, where annotation requirements remain high due to extensive pixel labeling. Extending ViT-based AL to the pixel-level, low-budget regime is uniquely challenging: with as little as one pixel query per image, a method must simultaneously identify the most informative locations and train a reliable decoder head from an extremely sparse signal. We introduce TEAL (Token-space Efficient Active Learning)…
My latest vibe coded game experiment is a Command & Conquer style RTS (Real Time Strategy) game called Battle Drones (play it here). Unlike my previous experiments, this one is actually fun to play. I'm somewhat embarrassed to admit that I got there by telling Claude to "Make it a LOT more fun". It also supports multiplayer between nearby devices using audio to synchronize game state! Here's what I've built so far, and some of the things I learned along the way. Battle Drones How I built this v1: the initial prototype v2: full sprites v3: varied terrain v4: the fun update v5: Battle Drones goes 3D v8: Archipelago v10: Goliath, and agent battles Agent battles v11: Random Maps v12: Multiplayer via sound Some thoughts Update: Online multiplayer support Battle Drones My initial goal was to bui
Vlad Mihalcea, Patrick Baumgartner, and Frank Delporte, three seasoned Java Champions, shared their experiences using GitHub Copilot in production environments and open-source projects, emphasizing human-led problem definition, context anchoring, iterative validation, and ultimate accountability—rather than relying on AI to automatically generate code.
MIT Technology Review, citing TripleTen’s 2026 research, notes that AI adoption across industries is not a matter of “early adopters versus laggards,” but rather follows a natural timeline of technological diffusion: technology and finance lead the way in validating toolchains and workflows, while compliance‑driven, human‑centric sectors such as healthcare and education, though currently at lower adoption rates (e.g., only 18% of healthcare organizations are required to use AI chatbots), are gaining a valuable window of opportunity to systematically identify high‑value skills—such as AI output validation and prompt engineering—and proactively build relevant capabilities.
We added GPT-6 Astra, Luna and Sol, Claude Opus 5.5 and Claude Sonnet 5.5 to the Agentic Analytics Benchmark: all five land mid-board within a point of each other, and Sonnet 5.5 brings the biggest change in turns, cost and truncated runs.
Release: llm-echo 0.5 I'm working towards a new LLM release, and testing PRs caused me to revisit the llm-echo debug plugin. This plugin provides a "model" which doesn't involve an LLM at all - it just repeats back the prompts that have been fed to it. The new release now updates the simulated token usage counts to take tool calls, system prompts, and reasoning tokens into account. Example usage: uvx --with llm-echo llm -m echo 'hi there' -o thinking 1 --usage Outputs: First I consider the prompt, then I decide what to say. { "prompt": "hi there", "system": "", "attachments": [], "stream": true, "previous": [] } Token usage: 2 input, 24 output Tags: llm
Release: web-edit 0.1a2 web-edit (previously) is my tiny web app for browsing and editing files in a folder. I mainly built it for editing files on different machines via Tailscale. I added a "copy to clipboard" option for both text and image data. I needed this because I was running pelican prompts on a laptop from my phone (here's DeepSeek 4.1 Flash 1017) and I wanted an easy way to copy the resulting logs into a Gist. Tags: projects, pelican-riding-a-bicycle, deepseek
This issue focuses on key developments in the AI field in late October 2026: Meta unveiled its multimodal MoE model, Muse; Spark Open has been met with skepticism and labeled “dead”; performance, licensing, and deployment ecosystems among China, the U.S., and Europe are becoming increasingly divergent; China plans to introduce a draft export control regulation covering 23 categories of AI technologies; and several cutting-edge models have been released, accompanied by unverified performance claims and security concerns.
We think "did a human read the code?" is an imprecise proxy for "what could go wrong with this change, and which check—automated or human—is the right one to catch it?" Asking that question also changed when review happens. Before, review was a gate at the end of the road: write the code, open a PR, wait for your colleague. Now, review runs continuously, more like spellcheck while you write. Most changes have been reviewed and fixed by Claude at least once before a human is asked to look. This catches most routine mistakes, and saves team attention for higher-level questions and scrutiny. — Addy Osmani, Morgan Westlee Lunt, and Qing Wang, How the Claude Code team does code review Tags: anthropic, claude, ai, claude-code, llms, addy-osmani, coding-agents, generative-ai, agentic-engineering,